摘要：AbstractThe paper discusses performance measures used dominantly in the robust and optimal control design. By simple examples of the first order time delayed (FOTD) system control it illustrates that the usually prescribed levels of the maximal and complementary sensitivity functions indeed define situations with interesting loop properties, but may not be universally applied to the robust and optimal design of systems with an uncertain feedback variable in a broader range. For this purpose the shape related measures based on deviations from monotonicity yield results matching the technological requirements of practice in a much more appropriate way. It is also shown that in an optimal nominal controller design the monotonicity based performance measures nearly coincide with the multiple real dominant pole (MRDP) method. For the loop optimization using a broader spectrum of different performance measures, the performance portrait method may be recommended. It avoids the problems of convergence to the absolute optimum and, once generated, the performance portrait may repeatedly be used with a limited effort for a broader spectrum of different cost function specifications.
